Canonical URL

One page, several front doors

The same page on your site is usually reachable at more than one address. With www and without it. With a slash on the end and without. With a tracking tag stuck on by whatever shared the link. A product that sits in two categories often has an address under each.

To you these are one page. To a search engine, every distinct address is a separate page until something tells it otherwise. The canonical is that something: a line in the page code that names the preferred address and asks the engine to credit that one.

Here is the everyday version. You paste your service page link into a post, the platform tacks a tracking string onto the end, a customer copies that version into a review, and a supplier links to the plain one without the www. Three addresses, one page, and the search engine now has to guess which of them is the page.

In practice it is one line in the code at the top of the page, holding the full address you picked, protocol and domain and all. Most pages point that line at themselves, which is exactly right and is the setting almost everyone should have.

It is a strong hint rather than a command. Engines usually honor a canonical, but they can pick a different address when the other signals on the site disagree with it.

What the split actually costs you

Credit gets divided. Links, internal or from other sites, are the currency search engines count, and when they land on three versions of one page, none of the three carries the full weight the page earned.

The wrong version can also be the one that shows. A results listing with a long tracking string trailing off the end reads as sloppy to the person deciding whether to click, and you did not choose it.

As a worked example, say a roofing company runs a spring promotion and shares the same metal roof page with three different tracking tags across three channels. Without a canonical, an engine can treat that as four pages saying identical things, then decide on its own which to show and which to ignore.

The other reason to care is that a wrong canonical does more harm than a missing one. A missing canonical leaves the engine guessing, and it usually guesses reasonably. A template that points every page on the site at the homepage is telling the engine, on every page, that none of them are the real page. That is a quiet way to remove a whole site from results, and it is the kind of bug that survives a redesign because nothing about the site looks broken.

Canonicals get set in the page template, not bolted on afterward, which is why they belong with the rest of your web development work.

Canonical or redirect, and how to choose

A 301 redirect sends both people and crawlers somewhere else and retires the old address. A canonical leaves both addresses working and simply names your preference. Use the redirect when an address should stop existing. Use the canonical when both need to keep working, like a product listed under two categories, or a page that keeps collecting UTM parameters.

Handled well, canonicals are the main defense against accidental duplicate content on your own site, and they steer which version ends up in the index.